Inherit ForwardRateStructure from ZeroYieldStructure

[eltoder]

Instead of duplicating its code. This is effectively a public API because users are encouraged to override zeroYieldImpl with an efficient implementation.

Also, delete dead forwardImpl in ZeroSpreadedTermStructure. This method is never called because the class is not a ForwardRateStructure.

[coveralls]

coverage: 73.815% (-0.001%) from 73.816%
when pulling 9579bed on eltoder:feature/fwd-zero-termstructure
into f5b93e4 on lballabio:master.

[eltoder]

@lballabio Thinking more about it, I think we should just deprecate ForwardRateStructure. It contains a very bad implementation of numerical integration, which it warns people not to use. All implementations of ForwardRateStructure in QuantLib already implement zeroYieldImpl. But if you implement zeroYieldImpl, your forwardImpl is no longer used. In this case it's better to inherit from ZeroYieldStructure and skip forwardImpl altogether. What do you think?

[lballabio]

Let me sit with this for a bit. I see your point though.

[github-actions]

This PR was automatically marked as stale because it has been open 60 days with no activity. Remove stale label or comment, or this will be closed in two weeks.

[eltoder]

@lballabio Did you have a chance to think about this?

[lballabio]

Ok, let's go ahead (in 1.41 — I've already cut 1.40). I'm not sure about deprecating ForwardRateStructure, but yes, we can skip the middleman and inherit classes such as InterpolatedForwardCurve directly from ZeroYieldStructure.